中文摘要：

轮廓检测是计算机视觉领域一个重要问题，目标和背景的纹理和噪声的存在让提取轮廓和保存轮廓完整性变得十分复杂。现有的神经生理学和解剖学研究成果，提供了许多轮廓检测的新方法。文中改进了现有的仿生模型，通过对图像进行不同参数的LoG滤波，并将结果赋予不同的权重，然后组合所有结果来抑制纹理，减弱了纹理和背景中的干扰成分对轮廓提取的影响，增强了目标的轮廓边缘，保留了目标轮廓的完整性。实验结果表明，这种方法有效地抑制了纹理，减少了对目标轮廓完整性的破坏，提高了图像的轮廓检测性能。

英文摘要：

Contour detection is a significant problem of computer vision field. The textures and noises of objects and backgrounds make it harder to get and keep a contour from an image. The anatomy and neurophysiology research results provide a lot of novel contour detec-tion method to solve this problem. In this paper,improve the existing bionic model,use a combination of image convoluted with LoG fil-ters and weight each result to inhibit the texture,weakening the texture and the background interference composition effects on contour ex-traction,enhancing the contour edge of object,making the contour integrated. The experimental results show that this method can inhibit the texture effectively and reduce the damage to the integrity of the target contour,strengthen the contour detection performance of image.